Filtering by
- All Subjects: GAN
- All Subjects: Temperature
- Creators: Wang, Yancheng
This paper is centered on the use of generative adversarial networks (GANs) to convert or generate RGB images from grayscale ones. The primary goal is to create sensible and colorful versions of a set of grayscale images by training a discriminator to recognize failed or generated images and training a generator to attempt to satisfy the discriminator. The network design is described in further detail below; however there are several potential issues that arise including the averaging of a color for certain images such that small details in an image are not assigned unique colors leading to a neutral blend. We attempt to mitigate this issue as much as possible.
This paper is centered on the use of generative adversarial networks (GANs) to convert or generate RGB images from grayscale ones. The primary goal is to create sensible and colorful versions of a set of grayscale images by training a discriminator to recognize failed or generated images and training a generator to attempt to satisfy the discriminator. The network design is described in further detail below; however there are several potential issues that arise including the averaging of a color for certain images such that small details in an image are not assigned unique colors leading to a neutral blend. We attempt to mitigate this issue as much as possible.
This paper is centered on the use of generative adversarial networks (GANs) to convert or generate RGB images from grayscale ones. The primary goal is to create sensible and colorful versions of a set of grayscale images by training a discriminator to recognize failed or generated images and training a generator to attempt to satisfy the discriminator. The network design is described in further detail below; however there are several potential issues that arise including the averaging of a color for certain images such that small details in an image are not assigned unique colors leading to a neutral blend. We attempt to mitigate this issue as much as possible.
In this thesis, an electro-thermal model of an AlGaN/GaN HEMT on a SiC substrate is simulated using Silvaco (Atlas) TCAD tools. Output characteristics, current density and heat flow at the GaN-SiC interface are key areas of analysis in this work. The electrical characteristics show a sharp drop in drain currents for higher drain voltages. Temperature profile across the device is observed. At the interface of GaN-SiC, there is a sharp drop in temperature indicating a thermal resistance at this interface. Adding to the existing heat in the device, this difference heat is reflected back into the device, further increasing the temperatures in the active region. Structural changes such as GaN micropits, were introduced at the GaN-SiC interface along the length of the device, to make the heat flow smooth rather than discontinuous. With changing dimensions of these micropits, various combinations were tried to reduce the temperature and enhance the device performance. These GaN micropits gave effective results by reducing heat in active region, by spreading out the heat on to the sides of the device rather than just concentrating right below the hot spot. It also helped by allowing a smooth flow of heat at the GaN-SiC interface. There was an increased peak current density in the active region of the device contributing to improved electrical characteristics. In the end, importance of thermal management in these high temperature devices is discussed along with future prospects and a conclusion of this thesis.